Lec 1: what this course is

A. why this course exists

  1. origin story: AI transformed how I teach and run 6.390; large variance in how quickly staff picked up AI skills; my own workflow could transfer rather easily; what failed to transfer was the habit of asking what needs AI
  2. the same gap shows up in UROP- or MEng-like research projects: a field report on eight agent-assisted scientific computing projects (2026) found the routine engineering broadly delegable while the researchers’ own time moved to defining what would count as correct
  3. and the split holds at the frontier: Co-Scientist (Nature, 2026) runs lit review, generation, and ranking at scale, with what to keep still decided outside the loop

B. wins and failures from my own use

  1. Anigans, my personal AI: scoped delegation, largely a win, and the reason is a build I own end to end
  2. gallery view of my other AI use
  3. common thread of all the wins: catered to my needs, guardrailed, designed with care; in short, I retained agency
  4. whereas the failures came when I went completely hands-off

C. judging “good”

  1. self-assessment misleads even experts: experienced developers working on their own repositories ran 19% slower while believing they were 20% faster (METR, 2025), so the feeling of speed is not evidence of speed
  2. similar “discrepancy” in learning, in a randomized four-group comparison (2024), essays revised with ChatGPT improved more than in the other three groups, while knowledge gain and transfer stayed flat, so the artifact improved and the learner did not
  3. elephant in the room: judging “good” can be subjective, so an anonymous poll on two or three cases, “is this good AI use for your learning?”, with two follow-ups, “did it help you finish?” and “could you now do the next one yourself?”; that second one asks students to rate their own learning, and such ratings track how the material felt, since they follow fluency even when memory does not (2012) and effort gets read as poor learning (2019), where students rated the more effortful study strategy less effective and mostly avoided it, while the ones who chose it scored higher on the later test than the ones who chose the easier strategy (with some nuances in reading these results)
  4. still, judgment vs. no judgment can be obvious; show the with/without design pair, same prompt, two pages, run as a second poll, “is this slop or polished”
  5. now, what counts as “good” is contested; whether judgment happened is plain; both were true before the AI era; so how did we learn what good judgment is, and how did we learn anything at all?

D. how we ever learned anything

  1. four eras of learning (oral, print, internet, AI); full circle, the medium is dialogue again, now with AI instead of a person
  2. and what has stayed the same? internalizing has always been the learner’s job, and judgment is what internalizing leaves behind
  3. trade poll: pick the era you would have wanted to learn in, then tell a neighbor one thing you would import from another era and one thing of your era you would give up to get it
  4. this “reading the present against history” is the lens we will apply all semester

E. the map and the thesis

  1. the semester at a glance: the module map, the two deliverables, how labs run
  2. every topic draws on EECS fundamentals, agentic AI tools, and applied ML projects; agency is the overlap
  3. what agency means here: owning the calls about what to delegate, what to verify, and what to keep for oneself; tell an agent to do something and even a slop setup will probably get it done, so whether the task got done cannot be the measure
  4. we argue that agency is the central skill for work with AI, the part that does not offload; students may disagree but should do so with evidence

Lab 1

  1. pair up; every pair gets the same default topic pair, engineering material from outside the prerequisite courses, each topic learnable in twenty minutes; the default pair is Kalman filters and coding theory, and a pair swaps to the staff-written alternate pair only if either partner already knows one of the defaults
  2. before studying anything, the two partners, A and B, each take the staff-written quiz: one part on Kalman filters, one part on coding theory; A and B record their scores, this first attempt is unscored and for calibration
  3. study: both partners study Kalman filters for the first twenty minutes, A with AI and B from print; then both study coding theory for the next twenty minutes, A from print and B with AI. Each student ends up having studied both topics
  4. write down the score you expect on each topic, then take the same quiz a second time, closed book, no notes and no AI
  5. each student reports, for each topic: the medium used, the expected score, the actual score
  6. before the results go up, one show of hands: which medium will show the bigger gap between expected and actual
  7. write three sentences for reflection: on which topic did you score better than you expected, on which worse, and does your result go the same way as the room’s pooled results or against it
  8. to close, compare strategies: staff draw at random among the high scorers willing to share, one per medium; the AI pick replays their conversation history, the print pick says what they did with the text, and others add what they tried
  9. write the week’s agency-log entry: one decision you made about using AI during this session, the options you had, what you chose, and how it turned out
  10. checkoff: walk a staff member through your reflection
  11. reading: Working Methods (Keith Thomas, 2010), a historian’s first-person account of pre-internet research practice
  12. optional: tag what the medium removed, on one staff-chosen 6.390 concept with both artifacts supplied: a fixed AI transcript, one question and its answer, and the 2015 route to the same concept, a search results page, two forum threads, and the textbook section a learner would have read; list the steps the 2015 route required that the transcript removes, tag each with the shared bank (question formation, reading around the answer, tolerating not-knowing, cross-checking, self-explanation, pure friction), and settle disagreements with a neighboring pair
  13. optional: argue the lecture’s four-era story (oral, print, internet, AI) is wrong or incomplete: propose a different cut of the history and name one thing it explains that the four eras miss

Page updated August 25, 2026.